Women and Sports 261

Home , Alicia Hollowell, Cat Osterman, Jennie Finch, National Pro Fastpitch, Pennsylvania Rebellion

09_BER_46412_ch08_259-290_PP3.indd 259

SAEED KHAN/Getty Images To answerthisquestion,Chapter8willexplore thefollowing: 1. 3. 2.

Women andSports for Women andMeninSports? Can MarketsExplainDifferences inOutcomes paid inwomen’s sports. exploring this gap, this chapter will discuss the process by which wages are sports, we also see such a gap in professional basketball. In addition to women andmenhasbeenwelldocumentedoutside of sports.Inside outcomes weobserve. sports highlight how much public policy and historical perceptions drive the sively onsupplyanddemandtoexplainoutcomesinamarket.Butwomen’s leadership abilities. idea, atestthatreveals noevidenceofmenandwomenhavingdifferent ter leadersthanwomen.The datainwomen’s sportsallowustotestthis hired toleadwomen’s teams.Thissuggeststhatmanybelieve men are bet- It isveryrare toseeawomanleadmen’s team.Incontrast, menare often and nations.Butinsports,womenonlyseemtoleadwomen’s sportsteams. Leadership inWomen’ Wages PaidinW The Demand for Women’s Sports: Not forredistribution. by WorthPublishers. ©2018 WorthPublishers.Distributed Copyright omen’s Sports:Thegapbetweenthewagespaidto s Sports:Women are leadersoffirms,universities, Traditionally, economists focus exclu- Chapter 8 8/31/17 1:04 PM 260 Sports Economics

Historically, sports have been played mostly by men. In 1971, 3.7 million boys played high school sports, while fewer than 300,000 girls did. The girls who played in high school had limited opportunities to continue doing so after graduation. In 1970, there were just 2.5 women’s teams per school in the National Colle- giate Athletic Association (NCAA); across all colleges, only about 16,000 women played sports.1 And for the few women who got to play college team sports, their career almost always ended with their college graduation. The professional team sports leagues we see today in women’s sports did not exist. Forty years later, these numbers are quite different. As of 2013, there were 4.49 million boys playing high school sports and 3.2 million girls. So while boys’ participation had increased by 21%, the participation rate among girls had increased by 966.7%. And upon high school graduation, many of these girls now find opportunities in college. In 2014, there were over 200,000 women playing college sports.2 When these women leave college, some will find opportunities to play professionally. Women are now paid to play in a variety of sports: the Wom- en’s National Basketball Association (WNBA), National Women’s Soccer League (NWSL), National Women’s Hockey League (NWHL), and National Pro Fastpitch (NPF). What explains these changes? The go-to explanation economists turn to is “market forces.” If we follow that story, then in 1970, very few women and girls were interested in playing sports. And soon afterward, demand for sports changed dramatically and suddenly many women and girls loved sports. Obviously, that story is somewhat ridiculous. A more plausible explanation begins with Title IX. Title IX is an amendment to the Civil Rights Act of 1964. Initially, the Civil Rights Act focused on discrimination with respect to race, color, religion, and national origin. But in 1972, the following was added: “No person in the United States shall, on the basis of sex, be excluded from participation in, be denied the benefits of, or be subjected to discrimination under any educational program or activity receiving federal financial assistance.”3 Although Title IX does not explicitly mention sports, its impact on sports is clear. Zimbalist (2001)4 noted that female participation in sports changed dramatically after the amendment was signed into law by President Richard Nixon. In 1971, 294,015 girls participated in high school sports. In 1973, this number

1R. Vivian Acosta and Linda J. Carpenter, “Women in Intercollegiate Sport: A Longitudinal, National Study, Thirty-Seven-Year Update, 1977‒2014.” Unpublished manuscript, 2014. Available at http://www.acostacarpenter.org. 2Acosta and Carpenter (2014). 3A discussion of the Civil Rights Act of 1964 and the specifics of Title IX can be found in Susan L. Averett and Sarah M. Estelle, “The Economics of Title IX Compliance in Intercollegiate Ath- letics, in Eva Marikova Leeds and Michael Leeds (eds.), Handbook on the Economics of Women in Sports (Cheltenham, UK: Edward Elgar, 2013). 4Andrew Zimbalist, Unpaid Professionals (Princeton, NJ: Princeton University Press, 2001).

09_BER_46412_ch08_259-290_PP3.indd 260 8/31/17 1:04 PM Women and Sports 261

increased to 817,073; by 1978, it was 2.08 million. Zimbalist (2001) observed a similar pattern at the college level. In 1971, only 31,852 women played college sports. By 1977, that number increased by more than 100% to 64,375. The story of Title IX indicates that it takes more than market forces to explain women’s sports. What appears to matter is the existence of opportunity. A brief story from the history of soccer further illustrates this point.5 There is evidence that women were playing soccer in the 19th century. In 1881, the Glasgow Herald reported on a match between teams of women from Scotland and England. In 1895, the British Ladies’ Football Club (soccer club) was founded. During World War I, women’s soccer in England took off. And the popularity of women’s soccer didn’t end when the war concluded in 1918. In 1920, 53,000 fans turned up to watch a women’s soccer match (with another 14,000 report- edly turned away). Such demand exceeded what was typical for men’s soccer at the time. Although market forces seemed to indicate that women’s soccer was eco- nomically viable, opportunity for this sport to grow was soon eliminated by non-market forces. In December 1921, the English Football Association (FA) declared that football was “quite unsuitable for women and not to be encour- aged.” Coaches and referees were told they would lose their licenses if women’s games were allowed on men’s fields. Yes, women’s soccer was banned. And that ban stayed in place in England until 1972.6 The story of Title IX illustrates how government can create opportunity. And the actions of the English FA in 1921 illustrate how a governing body can take away that opportunity. Both stories highlight that we need to do more than just appeal to “market forces” in explaining why something does or does not happen.

8.1 The Lesson Learned — and Not Learned — from Demand Data

In 2016, the average WNBA team saw 7,655 fans at each game.7 Meanwhile, the average National Basketball Association (NBA) team in 2015‒16 managed to attract 17,864 fans.8 The market has thus spoken. Women’s professional basketball is not as popular as men’s professional basketball.

5Amanda Coletta, “A League of Their Own: The Most Dominant Soccer Team in 1920 Was Full of Female Factory Workers,” The New York Times, June 5, 2015, http://nytlive.nytimes .com/womenintheworld/2015/06/05/a-league-of-their-own-the-most-dominant-soccer-team- in-1920-was-full-of-female-factory-workers/. 6Coletta (2015). 7http://www.wnba.com/news/record-breaking-attendance-five-years-digital-social-retail/. 8http://www.insidehoops.com/attendance.shtml.

09_BER_46412_ch08_259-290_PP3.indd 261 8/31/17 1:04 PM 262 Sports Economics

Once again, we need to work a little bit harder to draw a conclusion. Let’s start with a bit of history. The WNBA was founded in 1997. With the conclusion of the 2016 season, the league has only existed for 20 years. The NBA began as the Basket- ball Association of America (BAA) in 1946‒47.9 The league’s 20th season occurred in 1965‒66. That season the Boston Celtics, led by the league’s most valuable player (MVP) Bill Russell, won the NBA title. Wilt Chamberlain led the league in points, rebounds, and field goal percentage. NBA fans also were able to witness the talents of All-NBA players like Oscar Robertson, Jerry West, and Rick Barry. Despite this talent, though, the average team only drew 6,019 fans per game.10 To illustrate the popularity of the NBA around this time, in March 1962, Chamberlain scored 100 points in a game for the Philadelphia Warriors. This is a mark that has never been matched in NBA history. It was also a game that few people saw. The game took place in Hershey, Pennsylvania, nearly 100 miles outside of Philadelphia. The reported attendance was only 4,124.11 The NBA’s lack of popularity early in its history was not unusual. In baseball, the National League (NL) came into existence in 1876. In 1895, the average team only drew 3,690 fans per game.12 Six years later, the American League (AL) started to play. Twenty years into its history, the average AL team attracted 7,968 fans per game.13 A similar story can be told about the National Football League (NFL). In 1941, the average NFL team played before 20,157 fans, about 30% of the gate an average NFL team sees today.14 Early in a league’s history, attendance appears to be relatively low. Atten- dance is clearly not a function of only one factor.15 But one factor that does seem important early on is familiarity — how familiar the media and fans are with the teams and players in a league. Neale (1964)16 referred to this as the “fourth

9The NBA considers 1946 the year the league began. To illustrate, the 50th anniversary team was named in 1996. And three NBA franchises — the Knicks, Celtics, and Warriors — formed in the BAA in 1946. 10http://www.apbr.org/attendance.html. 11http://www.nytimes.com/packages/html/sports/year_in_sports/03.02.html. 12http://www.baseballchronology.com/Baseball/Years/1895/Attendance.asp. 13http://www.ballparksofbaseball.com/1920-29attendance.htm. 14https://sports.vice.com/en_us/article/how-the-wnba-compares-to-other-sports-leagues- at-age-20. 15Other issues we would consider is that population and incomes were lower in the past. That would depress attendance. Of course, there are many more entertainment options that would depress attendance today. And there is the possibility that women sports, in general, face cus- tomer discrimination. That being said, the history of sports suggests attendance in the early years of a league tends to be lower than what we see in later years. 16Walter Neale, “The Peculiar Economics of Professional Sports,” Quarterly Journal of Economics 78, no. 1 (1964): 1–14.

09_BER_46412_ch08_259-290_PP3.indd 262 8/31/17 1:04 PM Women and Sports 263

estate benefit.” Neale noted that one product of a sports league is coverage by the media. This essentially amounts to free advertising for the league. Early in a league’s history, however, such coverage might be relatively scarce. Consider once again Wilt Chamberlain’s 100-point game. There was no television broadcast of this game. And after the game, media coverage was less than impressive. Sports Illustrated reported the story in one paragraph on page 67 of the magazine. Daily newspapers did not see this as a front-page story.17 Again, at the time, the NBA — despite having existed for 18 years — was not considered a major sport by the media. Unfortunately, the media tends to have the same attitude toward all of women’s sports today. Cooky, Messner, and Musto (2015)18 reviewed how much time the sports media spends examining women’s sports. As Table 8.1 illustrates, historically, women’s sports have received very little coverage. Local network coverage in Los Angeles, via KCBS, KNBC, and KABC, devoted only 5% of its sports coverage to women’s sports in 1989. That percentage increased to 8.7% 10 years later. But after that, coverage of women’s sports declined, so in 2014 there was even less coverage of women’s sports than 25 years earlier. 19

Table 8.1 Coverage of Men and Women’s Sports by KCBS, KNBC, and KABC: 1989–201419 Year Men Women Neutral 1989 92.0% 5.0% 3.0% 1993 93.8% 5.1% 1.1% 1999 88.2% 8.7% 3.1% 2004 91.4% 6.3% 2.4% 2009 96.3% 1.6% 2.1% 2014 94.4% 3.2% 2.4%

Cooky et al. (2015) tell a similar story for ESPN Sportscenter in 2014. That year, only 2% of ESPN’s coverage was devoted to women’s sports. Perhaps such coverage simply reflects overall interest in women’s sports. But first, we need to understand that most sports coverage is done by men. The Women’s Media Center reported that, in 2015, only about 10% of sports reporters were women.20

17http://www.businessinsider.com/photos-in-1962-the-media-barely-noticed-that-wilt-chamberlain-scored-100-points-2012-3?op=1. 18Cheryl Cooky, Michael A. Messner and Michela Musto, “‘It’s Dude Time!’ A Quarter Cen- tury of Excluding Women’s Sports in Televised News and Highlight Shows,” Communication & Sport, 2015, pp. 1‒27. 19Cooky et al. (2015. p. 6). 20See Women’s Media Center, “The Status of Women in the U.S. Media 2015,” 2015, http:// wmc.3cdn.net/83bf6082a319460eb1_hsrm680x2.pdf.

09_BER_46412_ch08_259-290_PP3.indd 263 8/31/17 1:04 PM 264 Sports Economics

This percentage does not reflect the interest women have in sports. A 2015 Gallup poll indicated that 59% of adults consider themselves sports fans. Broken down by gender, 66% of men and 51% of women describe themselves as fans of sport.21 This means that about 43% of the population of sports fans are women. Nevertheless, 90% of those reporting on sports for the media are men.22 Maybe all those men are actually reporting on women’s sports as the market suggests they should. Unfortunately, the stories behind the sales of sponsorships and television rights suggest that the market doesn’t always reflect values per- fectly consistent with the data. The 2015 Women’s World Cup Final between the United States and Japan attracted 25.4 million viewers on Fox Television. An additional 1.3 million viewers watched the game in Spanish on Telemundo.23 The audience of 26.7 million viewers was larger than for any soccer match in U.S. history. The match attracted more viewers than those who had tuned in to see:

•• (in the United States) the 2014 Men’s World Cup Final between Germany and Argentina24 •• Game 7 of the 2014 World Series between the Kansas City Royals and San Francisco Giants25 •• the decisive Game 6 of the 2015 NBA Finals between the Golden State Warriors and Cleveland Cavaliers26

Despite the spectacular ratings, sponsorships for the broadcast of the Women’s World Cup seemed to be immensely undervalued. Whereas the Men’s World Cup brought in $529 million for ESPN in 2014, Fox was only paid a reported $17 million for the broadcast of the Women’s World Cup in 2015.27

21http://www.gallup.com/poll/183689/industry-grows-percentage-sports-fans-steady.aspx. 22The population of reporters is not the only issue. In 2016, Alison Overholt became the editor- in-chief of ESPN The Magazine. This was the first time a woman had been named editor of a major U.S. sports magazine. And yes, that means all the other editors of major sports magazines are men. 23http://www.nytimes.com/2015/07/07/sports/soccer/womens-world-cup-final-was-most- watched-soccer-game-in-united-states-history.html?_r=1. 24http://www.nytimes.com/2015/07/07/sports/soccer/womens-world-cup-final-was-most- watched-soccer-game-in-united-states-history.html?_r=1. 25http://www.nytimes.com/2015/07/07/sports/soccer/womens-world-cup-final-was-most- watched-soccer-game-in-united-states-history.html?_r=1. 26http://www.npr.org/sections/thetwo-way/2015/07/06/420514899/what-people-are-saying- about-the-u-s-women-s-world-cup-win. 27http://www.bloombergview.com/articles/2015-11-06/a-free-market-in-soccer-would- pay-women-more.

09_BER_46412_ch08_259-290_PP3.indd 264 8/31/17 1:04 PM Women and Sports 265

This disparity in sponsorship revenue does not seem consistent with the actual number of viewers. How is this possible? Sponsorship deals are made in advance of a broadcast and are based on forecasts made by those involved in the transaction. As Mike Mulvihill, a senior vice president for Fox Sports, noted after the 2015 Women’s World Cup: “No question, I underestimated where this would be. It’s one of the most pleasant surprises we’ve ever had.”28 The undervaluation of women’s sports does not just concern soccer. We see a similar story with respect to the WNBA. In 2013, the average WNBA broadcast attracted 231,000 viewers on ESPN and ESPN2. Meanwhile, the average Major League Soccer (MLS) broadcast — on the same networks — attracted 220,000 viewers. Such numbers would suggest the WNBA was the better investment. But the WNBA television deal with ESPN, signed in 2012, only pays the league $12 million per season. In contrast, in 2014, ESPN paid MLS $75 million per year to broadcast its games and Univision agreed to pay another $15 million.29 Can anything be done to change how the media covers sports? Back in 2013, only 7% of the sports media coverage in France was devoted to women’s sports.30 At that time, Valerie Fourneyron, the French minister of sport, decided to allocate 1 million euros per year to help media outlets increase their coverage of women’s sports. The government also mandated a day every year on which the sports media in France would exclusively focus on women’s sports. Three years after these efforts, the coverage of women’s sports in France has increased from 7% to 15%. This still does not reflect equality, but demonstrates substan- tial progress in a short period of time. It is easy to look at attendance numbers and broadcasting deals and argue that those statistics simply reflect the market. But both history and perceptions of gender in sports drive the numbers we see.31 And those numbers can be changed — for the better and worse — by non-market forces.

28http://www.nytimes.com/2015/07/07/sports/soccer/womens-world-cup-final-was-most- watched-soccer-game-in-united-states-history.html?_r=1. 29http://www.bloombergview.com/articles/2015-11-06/a-free-market-in-soccer-would- pay-women-more. 30http://www.sbs.com.au/topics/zela/article/2016/06/15/french-media-coverage-womens-sport- cause-celebre. 31Not only are women’s sports able to attract audiences that rival what we see for men’s sports, women are increasingly the fans watching men’s sports. It is reported that of the NFL’s 150 million fans, 45% are women. So women are not just supplying sports content, women are also often the consumers of this content (https://www.washingtonpost.com/business/ economy/women-are-pro-footballs-most-important-market-will-they-forgive-the-nfl/ 2014/09/12/d5ba8874-3a7f-11e4-9c9f-ebb47272e40e_story.html).

09_BER_46412_ch08_259-290_PP3.indd 265 8/31/17 1:04 PM 266 Sports Economics

8.2 The Gender Wage Gap in the WNBA

Francine Blau and Lawrence Kahn note that in 2014, women who worked full- time earned — on an annual basis — about 79% of what men earned.32 There thus appears to be a gender-based wage gap in the U.S. economy. In sports, we tend to see the same story repeat itself. The one clear exception is tennis. In tennis, women and men who win one of the four major33 champion- ships are awarded the same prize money.34 The person responsible for making this happen was Billie Jean King. In 1972, King won the U.S. Open Women’s Singles title and did not receive the same prize money as Ilie Năstase (who secured the Men’s Singles title). King responded by informing the U.S. Open that she would not play in 1973 if the prize money was not equal. The U.S. Open consequently opted to make the prizes equal, a practice eventually followed by other major sports. Two things should be noted about the process by which the gender wage gap was closed in tennis. First, women’s and men’s tennis are not that different in terms of popularity. A recent Harris poll revealed that women’s tennis fans out- number men’s tennis fans.35 And perhaps to illustrate this point, Sports Illustrated named Serena Williams its Sportsperson of the Year in 2015. But beyond the popularity of women’s tennis, we need to emphasize what Billie Jean King did in 1972. King didn’t just ask for equal wages. King announced she would stop playing until it happened. In essence, King threatened to go on strike to change the labor market outcome. As noted in our discussion of labor economics, until baseball players unionized and threatened to strike, the labor market in that sport also didn’t change. The power labor has to withhold its services can be illustrated by the outcomes observed when this doesn’t happen. Professional basketball has a sizable gender wage gap. The average NBA player in 2014‒15 was paid

32Francine Blau and Lawrence Kahn, “The Gender Wage Gap: Extent, Trends, and Explana- tion,” IZA Discussion Paper no. 956, January 2016. Paper to be published in the Journal of Economic Perspectives. It is important to note that this statistic does not mean that in every job women are paid 79% of what men make. And gender discrimination — although clearly part of the story — does not explain the entire difference. As Blau and Kahn (2016) indicated, approximately 38% of the gap may be attributed to discrimination. 33The four majors are the Australian Open, the French Open, Wimbledon, and the U.S. Open. 34This is not true in the non majors. In non majors, men and women are not always paid the same. 35http://espn.go.com/nba/story/_/id/14470482/michael-jordan-jordan-stays-atop-harris- poll-ahead-babe-ruth-muhamad-ali.

09_BER_46412_ch08_259-290_PP3.indd 266 8/31/17 1:04 PM Women and Sports 267

$4.45 million.36 In contrast, the average WNBA player was paid $75,000.37 So these numbers suggest that women earned 1.7% of what men were paid to play professional basketball. Obviously, such a comparison is misleading. There is a significant revenue gap between the NBA and WNBA. According to Forbes.com, in 2013‒14, the NBA netted $4.7 billion in revenue.38 Forbes.com does not report similar data for the WNBA, but Harris and Berri (2016)39 offers an estimate of revenues from other sources. For example, we have already noted that the WNBA has a television contract that pays it $12 million per season.40 We also know average attendance in 2014 was 7,578 per game41 and that the average ticket price is at least $15.42 With 204 home games, gate revenue for the 2014 season was at least $23.2 million. Of course, there are other sources of revenue (merchandise, sponsorships, playoff ticket sales, etc.). But we can estimate that WNBA revenues, as Table 8.2 details, were a minimum of $35 million in 2014.43

Table 8.2 Estimate of Women’s National Basketball Association Revenues: 201443 Revenue Factors Revenue Television revenue $12,000,000 Average attendance 7,578 Average ticket price $15 Gate revenue per game $113,670 Total gate revenue for 204 regular-season games $23,188,680 Total revenue $35,188,680

36According to basketball-reference.com, total team payroll in 2014‒15 was $2,190,680. There were 492 players logging minutes that season, so the average salary was $4,554,603. 37http://sportsday.dallasnews.com/dallas-mavericks/mavericksheadlines/2015/07/26/ sefko-why-wnba-has-never-been-stronger-as-league-enters-dallas-market. 38http://www.forbes.com/nba-valuations/list/. 39Jill Harris and David Berri, “If You Can’t Pay Them, Play Them: Fan Preference and Own- Race Bias in the WNBA,” International Journal of Sport Finance 11 (August 2016): 163‒180. 40http://www.sportsbusinessdaily.com/Daily/Issues/2013/03/28/Media/WNBA.aspx. 41http://www.sportsbusinessdaily.com/Journal/Issues/2015/09/21/Leagues-and-Governing- Bodies/WNBA.aspx. 42This was reported for 2011 in David J. Berri and Anthony Krautmann, “Understanding the WNBA On and Off the Court,” in Eva Marikova Leeds and Michael Leeds (eds.), Handbook on the Economics of Women in Sports (Northampton, MA: Edward Elgar, 2013), pp. 132‒155. 43Table originally appeared in Harris and Berri (2014, p. 249).

09_BER_46412_ch08_259-290_PP3.indd 267 8/31/17 1:04 PM 268 Sports Economics

We have noted that the average salary in the WNBA is $75,000. There were 154 women who logged minutes in the league in 2014. So the WNBA paid $11,550,000 to its players in 2015. This means that the television deal was enough to pay every single WNBA player. The NBA pays 50% of its revenue to its players. The above analysis, which underestimates league revenue, says the WNBA only pays 32.8% of its revenue to its players. And that means, WNBA players are only receiving about 70% of what NBA players would be paid. There is additional evidence that the women in the WNBA are underpaid. The WNBA reports that in 2014, 101 women in the league supplemented their income by playing in women’s basketball leagues in other countries.44 The wages paid to the WNBA stars to play in these other leagues often dwarf what the WNBA offers. For example, Brittney Griner was paid $600,000 to play in the Women’s Chinese Basketball Association in 2014. It was also reported that both Sylvia Fowles and Maya Moore were paid at least $600,000 to play in the same league.45 Meanwhile, Diana Taurasi was paid $1.5 million by a team in the Russian Premier League and an additional sum — enough to at least cover her WNBA salary — to sit out the 2015 WNBA season.46 Yes, Taurasi was paid to not play in the WNBA. The Taurasi story suggests salaries in the WNBA are too low. And it doesn’t appear that this will change soon. The current labor agreement between the WNBA and its players calls for a $121,500 cap on the maximum salary in the league in 2021.47 What would that number be if the stars of the WNBA were paid like those of the NBA? Again, the NBA pays 50% of its salaries to its players. Let’s imagine that is the only restriction facing WNBA players. And let’s argue that players are paid only for wins. In 2015, league MVP, Elena Delle Donne, produced 8.3 wins for the Chicago Sky.48 This mark led the league and represented 4.1% of all wins in

44http://www.wnba.com/archive/wnba/news/2014_overseas.html. 45http://www.businessinsider.com/brittney-griner-basketball-china-2014-4. 46http://espn.go.com/wnba/story/_/id/12272036/diana-taurasi-decision-sit-spark-wnba- salary-changes. 47The current maximum salary is $107,500 (https://sports.vice.com/en_us/article/basketballs- gender-wage-gap-is-even-worse-than-you-think). For the future value, see http://wnbpa- uploads.s3.amazonaws.com/docs/WNBA%20CBA%202014-2021Final.pdf. 48The calculation of Wins Produced follows the method described in Berri and Krautmann (2013), which followed earlier work reported in David J. Berri and Martin B. Schmidt, Stumbling on Wins: Two Economists Explore the Pitfalls on the Road to Victory in Professional Sports (Princeton, NJ: Financial Times Press, 2010), and David J. Berri, “A Simple Measure of Worker Productivity in the National Basketball Association,” in. Brad Humphreys and Dennis Howard (eds.), The Business of Sport, 3 vols. (Westport, CT: Praeger, 2008), pp. 1‒40.

09_BER_46412_ch08_259-290_PP3.indd 268 8/31/17 1:04 PM Women and Sports 269

2015. Following the same approach documented in Table 8.1, one can estimate league revenues in 2015 at $34.4 million.49 If the players were paid 50% of the $34.4 million in league revenue, then Delle Donne was worth $700,754 in 2015, or nearly seven times the league’s salary cap. The NBA also has a cap on individual salaries. It stipulates that a player with less than seven years of experience can only be paid 25% of the league’s cap.50 Given a league payroll cap of $17.2 million going to players, each of the 12 teams could only pay $1,433,045 to its members. And 25% of that is $358,261. This number is about half of what we cited above, but still more than three times the salary Delle Donne was, in fact, paid in 2015. Why are WNBA players paid so little? One issue may be that even though the WNBA has a union, it has never come close to threatening a strike. Again, the history of labor relations in sports suggests that outcomes improve for labor once labor is able to credibly threaten to withhold its services. Two recent examples highlight the power of threatening to withhold labor. In 2015, the women’s national soccer team in Australia, the Matildas, had advanced further in the World Cup than any of the men’s teams in that nation’s history. Despite their success, however, the Matildas were only paid $500 (Aus- tralian dollars) in match fees, while members of the men’s soccer team received $7,500 (Australian dollars). The disparities in pay and financial support led the players on this team to skip a tour of the United States after the World Cup. 51 As a result of this strike, the members of the Australian women’s soccer team did see their pay increase.52 A similar story took place with respect to the United States Women’s National Hockey Team. In 2005, this team won its first gold medal in the International Hockey Federation Women’s World Championship. This team then proceeded to win the gold medal again in 2008, 2009, 2011, 2013, and 2015.53 Despite this success, the women who competed on these hockey teams were not well compensated. The women of the national team were only paid $6,000 every

49Television revenue was still $12 million in 2015. Average attendance was 7,318. If the average ticket price was still $15, then gate revenue was $22,393,080. And that means league revenue was $34,393,080. 50http://www.cbafaq.com/salarycap.htm#Q18. 51http://thinkprogress.org/sports/2015/09/10/3699819/heres-why-the-australian-womens- soccer-team-is-on-strike/. 52http://www.smh.com.au/sport/soccer/matildas-strike-vindicated-by-pay-increase-says- midfielder-hayley-raso-20151109-gkucdk.html. 53All-time tournament results can be found here: http://www.usawomenshockey.com/page/ show/1429977-year-by-year-results

09_BER_46412_ch08_259-290_PP3.indd 269 8/31/17 1:04 PM 270 Sports Economics

four years by USA Hockey.54 Essentially, they received a small stipend to train for the Olympics. But they were asked to compete in tournaments outside the Olympics. And for those there was no compensation. Prior to the 2017 World Championship, though, the women threatened to go on strike. USA Hockey responded by first trying to find replacement players.55 But when this approach failed, USA Hockey turned to nego- tiations. In the end, the deal struck dramatically changed wages for these women. According to an ESPN.com report, annual pay for the women on the national team was increased to approximately $70,000.56 So over a four year period, a woman on this team would receive about $280,000, or a 4,667% increase in pay. Increasing athlete pay was not the only result of this strike. But the dramatic change in pay alone highlights the power of the labor strike. As we have noted, fans and the sports media frequently condemn labor disputes. However without such disputes, we often see inferior labor market outcomes for athletes in professional sports.

8.3 The Highest-Paid Women in Professional Team Sports in North America

We have seen that WNBA stars can command much higher wages outside of North America. Consequently, WNBA stars are not the highest-paid women in professional team sports in North America. That honor currently goes to Monica Abbott, a pitcher in the National Pro Fastpitch (NPF) league. In 2016, she agreed to a six-year contract with the Scrap Yard Dawgs (an NPF expansion team) worth $1 million.57 Why was Abbott paid so well? To begin with, she may be the greatest pitcher in NPF history. As the discussion of how to measure an NPF pitcher’s performance indicates, Abbott — along with Cat Osterman and Jennie Finch — ranks

54see http://www.espn.com/espnw/voices/article/18908360/time-usa-hockey-wake-support- women-team 55http://www.espn.com/espnw/sports/article/19004695/usa-hockey-reaches-d-iii-players-rec- league-players-potential-replacements 56http://www.espn.com/olympics/story/_/id/19026627/usa-hockey-us-women-national-team- reach-agreement-avoid-boycott 57Abbott’s contract calls for her to receive a base salary of $20,000 per year, with attendance bonuses adding $880,000 to her total six-year contract (http://espn.go.com/espnw/sports/ article/15464430/pitcher-monica-abbott-signs-1-million-contract-national-pro-fastpitch- expansion-team).

09_BER_46412_ch08_259-290_PP3.indd 270 8/31/17 1:04 PM Women and Sports 271

among the top three pitchers in league history. But Abbott’s pay is not just a function of her skill. After all, Delle Donne and the other stars of the WNBA are also amazingly skilled basketball players. There also seems to be an issue with the market for Abbott’s services. First of all, Abbott was a free agent when she negotiated this contract. According to the NPF website, “Players who are not currently under contract and have not been drafted are considered free agents. . . .”58 Abbott’s contract had expired with the Chicago Bandits (her 2015 team), so she was free to sign with the Scrap Yard Dawgs. But why was this team willing to pay so much to sign her? A few factors were cited at the time of signing. The owner of the Scrap Yard Dawgs, Connie May, noted that the team was an expansion team and they wished to sign a player who would bring them immediate publicity. In other words, the Scrap Yard Dawgs had significant demand at this time for a star to give the team some needed media attention. There was also the issue of Abbott’s other employment choices. The average college graduate has a starting salary of $50,000.59 According to the NPF website, the average salary in the league is only $5,000 to $6,000. Since players are typically college graduates, they often leave the NPF to accept jobs outside of professional athletics. In addition, many NPF players play professionally in Japan, where they make more money than the NPF can offer. In sum, Abbott had other opportunities that would pay her more. Abbott — as the discussion of measuring pitching performance illustrated — is also a unique talent. Only Cat Osterman and Jennie Finich rival Abbott’s skill, and those players have retired. As Connie May noted: “We don’t have another Monica, Cat Osterman or Jennie Finch coming up in the foreseeable future. So if we miss this opportunity, then it’s on us.”60 Abbott can thus command a million dollar salary, while other NPF talents are paid considerably less. The story of Abbott’s contract highlights the importance of bargaining power in setting wages. On the demand side, the Scrap Yard Dawgs needed a major talent to attract publicity. The talent they wanted had other options and was unique in the market. All this shifted bargaining power to Abbott and allowed her to command a wage well beyond what other women in professional team sports in North America can.

58http://www.profastpitch.com/home/. 59http://time.com/money/3829776/heres-what-the-average-grad-makes-right-out-of-college/. 60http://www.foxsports.com/mlb/story/monica-abbott-million-dollar-deal-softball-landmark- womens-sports-050516.

09_BER_46412_ch08_259-290_PP3.indd 271 8/31/17 1:04 PM 272 Sports Economics

History of the NPF The 2015 Women’s College World Series (WCWS) attracted 31% more television viewers than the Men’s College World Series, with 1,196,000 average viewers.61 Despite the relatively greater popularity of the WCWS, its participants who become professional softball players do not tend to experience the same outcome as the men who eventually sign on as Major League Baseball (MLB) players. As noted in previous chapters, MLB has existed for well over a century and, in 2015, more than 73 million fans attended MLB games. In addition, MLB games are broadcast on ESPN, Fox, and TBS. So although fewer fans see the men who play college baseball, many will watch the few college baseball players who eventually make a major league roster. In contrast, women who become professional softball players in the United States graduate to the NPF league, which has been around since 2004. Like most new sports leagues, teams have entered, exited, and/or relocated. In all, 14 different franchises have played in the league, with no more than 7 playing in a given season. In 2015, those teams were the Akron Racers, Chicago Bandits, Dallas Charge, Pennsylvania Rebellion, and USSSA Pride (with their home field in Kissimmee, Florida). In 2016, the Scrap Yard Dawgs were added (with their home field in The Woodlands, Texas, just north of Houston). Of these teams, only Akron has existed since 2004. Chicago entered the league in 2005, while USSSA Pride began play in 2009. The remaining franchises all started in the last three seasons. According to the league website, the league attracts between 1,500 and 2,500 fans per game.62 In 2015, the 5-team league played a 48-game schedule, or had 120 home games. That means in 2015, the league attracted 240,000 fans for the entire season, or the approximate number of spectators a MLB team draws in about 8 games. Of course, we need to put those attendance numbers in proper perspective. The NPF began its 13th season in 2016. In 1890, the NL was in its 15th season. That year, the league attracted 776,042 fans to the ballpark. With 8 teams and 66 home games per squad, the league averaged 1,469 fans per contest,63 similar numbers to what the NPF draws today. One should note that comparing attendance in 1890 in the NL to what we see today in the NPF is problematic. As noted earlier in this chapter, populations in the past were much smaller. But also we noted, there are now far more entertainment

61https://nfca.org/index.php?option=com_content&view=article&id=6281:record-viewer- ship-at-the-2015-women-s-college-world-series&catid=109&Itemid=149. 62http://www.profastpitch.com/about/faqs/. 63http://www.ballparksofbaseball.com/1890-99attendance.htm.

09_BER_46412_ch08_259-290_PP3.indd 272 8/31/17 1:04 PM Women and Sports 273

options. So there are forces today that would make attendance larger (i.e. more population) and smaller (i.e. more entertainment options). The point made with this comparison is to highlight again how few people attended what we consider the major professional sports leagues when these leagues were in their first few decades of existence. The poor attendance we saw in the past in men’s professional sports highlights how long it takes to build demand in a professional sports league.

Evaluating Pitchers in the NPF There is a sense that when it comes to player statistics, baseball’s numbers are the best at evaluating the performance of individual players. Although this might be argued for hitters in baseball,64 it certainly can’t be argued for pitchers. The standard metric for pitchers is earned run average (ERA). But as McCracken (2001) argued65 — and as we noted in Chapter 6 — pitchers and the defensive players often work together to get the opponent out. As McCracken noted, ERA is a poor measure of a pitcher’s performance. A better approach is to focus on defensive independent statistics like walks, strikeouts, home runs, and hit batsmen. In other words, we should ignore statistics attributed to pitchers who partially depend on the actions of others. Obviously, this list begins with wins and losses since pitchers cannot win or lose a game by themselves. But it also includes hits allowed, sacrifice hits, sacrifice flies, and earned runs. All these partially depend on the action of fielders. McCracken illustrated the issue by noting the consistency of a pitcher’s performance with respect to these statistics. Bradbury (2007a)66 highlighted McCracken’s point by looking at the year-to-year correlation for major league hitters for strikeouts, walks, hit-by-pitch, home runs, ERA, and batting average allowed on balls in play (BABIP).67 Bradbury (2007a) reported the following correlations:

•• Strikeouts: 0.78 •• Walks: 0.64 •• Hit-by-pitch: 0.51

64Berri and Schmidt (2010) note problems with this argument for hitters. 65Voros McCracken, “Pitching and Defense: How Much Control Do Hurlers Have,” January 23, 2001, http://www.baseballprospectus.com/article.php?articleid=878. 66J. C. Bradbury, “Does the Baseball Labor Market Properly Value Pitchers?,” Journal of Sports Economics 8, no. 6 (2007a): 616‒632. 67BABIP = (Hits – Home Runs) / (At Bats – Strikeouts – Home Runs + Sacrifice Flies). This simply measures how often batted balls that are not home runs actually become hits (http:// www.fangraphs.com/library/pitching/babip/).

09_BER_46412_ch08_259-290_PP3.indd 273 8/31/17 1:04 PM 274 Sports Economics

•• Home runs: 0.47 •• ERA: 0.35 •• BABIP: 0.25 As McCracken noted, the factors that are independent of the defense around the pitcher are more consistent than those that rely partially on the performance of a team’s defense. This led McCracken to develop the defense independent pitching statistics earned run average (DIPS ERA) or what Baseball Prospectus refers to as fielding independent pitching (FIP).68 Bradbury (2007b) presented the basic methodology.69 A pitcher’s ERA is regressed on the FIP factors (strikeouts, walks, hit-by-pitch, home runs) and BABIP.70 And the results create a more stable measure of a pitcher’s performance.71 This basic idea has been used to create a FIP measure for the NPF. The data employed included all pitchers in the NPF who pitched at least 50 innings from 2004 to 2016. The model involved regressing each pitcher’s ERA on the defensive independent statistics (again, strikeouts, walks, hit-by-pitch, and home runs) and a defensive dependent measure (hits per ball in play or HperBIP).72 The results, reported in Appendix A of this chapter, indicate that 89% of the variation in a pitcher’s ERA may be explained by these factors. To construct a FIP ERA measure for pitchers, we first multiply each player’s defensive independent statistic by the corresponding coefficient. We then use the average value for HperBIP in the sample to construct each player’s FIP ERA. An example will help illustrate the process. The following table reports Monica Abbott’s defensive independent statistics for 2016. In addition, the estimated coefficient from our model is reported.

68See http://www.baseballprospectus.com/glossary/index.php?search=FIP. As acronyms go, FIP sounds much better than DIPS. So, as we noted in Chapter Six, we will follow the lead of Baseball Prospectus and call this FIP. 69J. C. Bradbury, The Baseball Economist: The Real Game Exposed (New York: Dutton, 2007b). 70Bradbury (2007b) also controlled for age of the pitcher, league of the pitcher, and season played. His model explains 77% of the variation in a pitcher’s ERA. 71In Hot Stove Economics: Understanding Baseball’s Second Season (New York: Copernicus, 2010), Bradbury notes in a different study of pitchers that ERA has a 0.30 year-to-year correlation, while DIPS ERA has a 0.54 year-to-year correlation. 72BABIP, which Bradbury used, requires sacrifice flies. We do not have sacrifice flies for every year in the NPF. Consequently, we turn to HperBIP, or [Hits – Home Runs]/[Outs + Hits – Strikeouts – Home Runs]. This metric was originally used by McCracken; see http://sabr.org/ research/many-flavors-dips-history-and-overview. Additionally, the model included team- specific fixed effects.

09_BER_46412_ch08_259-290_PP3.indd 274 8/31/17 1:04 PM Women and Sports 275

To calculate FIP ERA, we multiply the pitcher’s statistic by the corresponding coefficient. We next sum the values. And then we add to this sum both the constant term and the impact of the HperBIP (i.e., the league average for HperBIP multiplied by its corresponding coefficient). The result is the pitcher’s FIP ERA.

Value for Defensive Independent Estimated Column 2 3 Monica Abbott, Statistics Coefficient Column 3 2016 Strikeouts per 7 innings pitched 9.077 20.123 21.118 Walks per 7 innings pitched 1.619 0.195 0.315 Hit-by-pitch per 7 innings pitched 0.442 0.478 0.211 Home runs per 7 innings pitched 0.343 1.623 0.558 Defensive Dependent NPF League Statistics Value NFP average hits per ball in play (per 7 innings pitched) 0.271 12.729 3.453 Constant 21.916 FIP ERA 1.502 Actual ERA 0.981

As one can see, Abbott’s actual ERA in 2016 was 0.981, a mark that led the league. Her FIP ERA was a bit higher, but her mark of 1.502 also led the league. To put Abbott’s mark in perspective, the average ERA across the entire sample was 2.55. The average FIP ERA is also 2.55. Thus, Abbott in 2016 was much better than average. But as Table 8.3 indicates, it was not her best mark or even among the top 10 marks in league history. Topping this list was Jennie Finch in 2007, who had a 0.105 ERA and a 0.302 FIP ERA. Both marks are the best measures — among pitchers with 50 innings pitched — in league history. How good was Finch? Although her 2007 FIP ERA set the league record, her career FIP ERA was only 1.51. Yes, this is amazing. But both Cat Osterman (with a career mark of 1.35) and Abbott (with a career market of 1.25) were a bit better. In sum, the trio of Finch, Osterman, and Abbott makes for the three best pitchers in league history. Although these three were the best, it was possible for NPF hitters to actually hit off their league’s pitching. The same cannot be said of MLB players. In 2003, 2004, and 2005, Finch participated in a tour where she faced such legendary hitters as Albert Pujols, Barry Bonds, and Alex Rodriguez. None of

09_BER_46412_ch08_259-290_PP3.indd 275 8/31/17 1:04 PM 276 Sports Economics

Table 8.3 Top 10 National Pro Fastpitch Seasons, by Pitching, in League History: 2004–15 NPF Pitcher Year Team ERA FIP ERA Jennie Finch 2007 Chicago Bandits 0.105 0.302 Cat Osterman 2009 Rockford Thunder 0.415 0.749 Monica Abbott 2015 Chicago Bandits 0.310 0.904 Monica Abbott 2011 Chicago Bandits 0.802 0.985 Monica Abbott 2013 Chicago Bandits 0.925 1.069 Monica Abbott 2010 Tennessee Diamonds 0.789 1.166 Alicia Hollowell 2007 Akron Racers 1.194 1.188 Lauren Bay 2005 Chicago Bandits 0.883 1.189 Jamie Southern 2005 Akron Racers 0.750 1.192 Christa Williams 2005 Texas Thunder 0.756 1.194

these hitters could get a hit off of Finch’s pitches. In fact, Bonds, the best hit- ter at the time, could only hit the ball if Finch told him exactly which pitches were coming. And by “hit,” all Bonds could do was “tap a meek foul ball a few feet.”73 The same cannot be said for the women of the NPF. Yes, Finch was dominant. But women did get actual hits — and even home runs — off of her. So when you are watching NPF softball, remember: You are seeing women do something that major league hitters cannot!

8.4 Are Men Really Better Leaders?

Women make up roughly 50% of the population. But women tend to occupy only 20% of leadership positions. Such a result may be seen across a wide variety of industries and government positions. For example,74 in 2013, women held:

•• 24.5% of leadership positions in academia •• 23.5% of leadership positions in the arts and entertainment sector

73http://www.si.com/more-sports/2013/07/24/sports-gene-excerpt. 74The following derives from “Benchmarking Women’s Leadership in the United States,” a report prepared by the University of Denver, Colorado Women’s College, in 2013 (http:// www.womenscollege.du.edu/media/documents/BenchmarkingWomensLeadershipintheUS.pdf).

09_BER_46412_ch08_259-290_PP3.indd 276 8/31/17 1:04 PM Women and Sports 277

•• 15% of executive positions in Fortune 500 businesses •• 10% of CEO positions at the top 10 banking companies •• 23.3% of leadership positions in journalism and media •• 23% of leadership positions in law •• 25.5% of leadership positions among medical school faculty, regulatory agencies, and public and private hospitals •• 20% of all leadership positions in the technology sector •• 22.8% of all political and governmental leadership roles •• 18% of seats in Congress

These percentages are much lower in men’s professional and college sports. In fact, it is extremely rare for a woman to coach a men’s sports team in North America. So both within and outside of sports, men are clearly favored for leadership roles. This suggests that people believe men are better leaders. Is there any empirical evidence, though, to support such a contention? Let’s start with college coaching. A recent study indicates that there are no female coaches of any male teams in all of Division I sports.75 In contrast, today, male coaches are quite common in women’s college sports, with males holding more than 55% of head coaching jobs.76 There thus appears to be some disparity in the market for head coaching jobs in college sports. According to Acosta and Carpenter, after Title IX was passed in 1972, more than 90% of women’s teams were coached by women. But by 1978, that percentage had fallen to 58.2%. And since then, as Table 8.4 illustrates, the percentage has gradually declined. Today, less than 44% of women’s teams are coached by women. Once again, a similar table for men’s sports would reveal that women are essentially shut out from coaching men’s teams. So men are perceived to be qualified to coach women, but women are not perceived to be qualified to coach men. The sports of softball and baseball highlight this difference. Softball is played by women and baseball is played by men. According to Acosta and Carpenter, 83.5% of softball teams were coached by women in 1977. In 2014,

75Kate Fagan and Luke Cyphers, “The Glass Wall: Women Continue to Shatter Stereotypes as Athletes. So How Come They Can’t Catch a Break as Coaches?,” ESPN The Magazine, 2015, http://sports.espn.go.com/espn/eticket/story?page=theGlassWall. 76Acosta and Carpenter (2014).

09_BER_46412_ch08_259-290_PP3.indd 277 8/31/17 1:04 PM 278 Sports Economics

Table 8.4 Percentage of Women Coaching National Collegiate Athletic Association Women’s Teams77 Percentage of Women Percentage of Women Year Year Coaching Women’s Teams Coaching Women’s Teams 2014 43.4% 1994 49.4% 2012 42.9% 1992 48.3% 2010 42.6% 1990 47.3% 2008 42.8% 1988 48.3% 2006 42.4% 1986 50.6% 2004 44.1% 1984 53.8% 2002 44.0% 1982 52.4% 2000 45.6% 1980 54.2% 1998 47.4% 1978 58.2% 1996 47.7% 1972 90%1

this percentage had dropped to 66.3%. Meanwhile, women do not coach any NCAA baseball teams. Given the nature of the sports, though, this pattern might seem odd. The men who coach softball did not play softball in college. Again, this is a sport played by women. And yet, men, who presumably only have a background in baseball, are considered qualified to coach women’s softball. But women, who only have a background in softball, are apparently not qualified to coach baseball.77 Perhaps — as the aforementioned non-sports data seem to suggest — men are just “better” leaders. To address this issue, von Allmen (2013) examined how the gender of coaches affected outcomes in college softball.78 The specific model von Allmen employed incorporated the dependent and independent variables listed in Table 8.5 (complete details of this model are reported in Appendix A of this chapter). The key variable in this model is the dummy variable for gender. And Table 8.6 reports the estimated link between the gender of the coach and outcomes in college softball. As noted in the table, gender is not statistically significant. So it does not appear men are better than women at coaching softball.

77Acosta and Carpenter (2014). 78Peter von Allmen, “Coaching Women and Women Coaching: Pay Differentials in the Title IX Era,” in Eva Marikova Leeds and Michael Leeds (eds.), Handbook on the Economics of Women in Sports (Northampton, MA: Edward Elgar, 2013), pp. 269‒289.

09_BER_46412_ch08_259-290_PP3.indd 278 8/31/17 1:04 PM Women and Sports 279

Table 8.5 Evaluating Gender and Coaching in College Softball Dependent Variable RPI: Rank of team according to the rating percentage index (an index that considers both winning percentage and strength of schedule) Independent Variables Gender: dummy variable for gender of coach (1 = female, 0 = male) ShareExp: ratio of expenditures for each program to the expenses of all programs in the sample PAC-10: dummy variable for teams in the PAC-10, the dominant conference in softball WS10: number of World Series appearances over the 10-year period, 1996–2005

79 Table 8.6 Link Between Gender of Coach and Results in College Softball79 Variable Coefficient Standard Error t-Statistic p-Value Gender 4.14 5.17 0.80 0.425

von Allmen (2013) also looked at team outcomes given team history and expenditure. A different approach is to directly examine how a coach affects the performance of individual workers.80 Specifically, we will discuss what impacts the performance of a college women’s basketball player. For example, ESPN’s website offers an abundance of data on men’s college basketball. This includes box score statistics for every single player in NCAA Division I. But ESPN.com does not provide box score statistics for individual women playing college basketball. One can find such data at stats.NCAA.org, but complete data only exists back to the 2012–13 season (data for men’s college basketball players goes much further back. However, since there are more than 300 women college teams in each year, there is more than enough data to conduct a study. The study will employ the model detailed in Table 8.7. The dependent variable is the productivity of a woman in the current season per 40 minutes. This is regressed on a collection of independent variables: year in school, games played (to control for injury), productivity of teammates, position played, team expenditures per player, and past performance. Our focus, however, is the impact

79von Allmen (2013, p. 281). 80This study that follows is based on the work of Lindsey Darvin, Ann Pegoraro, and David Berri, “The Head Coach Role — Is It Only a Job for Men? An investigation of Head Coach Gender and Player Performance in the WNBA and NCAA Women’s Basketball,” working paper, 2016 followed the approach taken by David J. Berri, Michael Leeds, Eva Marikova Leeds, and Michael Mondello in “The Role of Managers in Team Performance,” International Journal of Sport Finance 4, no. 2 (May 2009): 75‒93, which looked at how player performance is impacted by coaches in the NBA.

09_BER_46412_ch08_259-290_PP3.indd 279 8/31/17 1:04 PM 280 Sports Economics

Table 8.7 Evaluating Gender and Coaching in National Collegiate Athletic Association Women’s Basketball Dependent Variable P40: Wins’ productivity per 40 minutes or a player’s overall productivity per 40 minutes played81 Independent Variables Gender: dummy variable for gender of coach (1 5 female, 0 5 male) ClassID: equal to 1 if the player is a freshman, 2 if sophomore, etc. GP: games played last two seasons (to control for injury) TMPROD: productivity of players on team DBIG: dummy variable equal to 1 if the player is a power forward or center DGUARD: dummy variable equal to 1 if the player is a point guard or shooting guard TOTEXPPER: total expenditure on program per player LagP40: player productivity the previous season

of coaches’ gender. And as one can see in Table 8.8, it appears that the gender of the coach does not impact the performance of individual players. 8182

Table 8.8 Link Between Gender of Coach and Individual Player Performance in National Collegiate Athletic Association Women’s Basketball82 Coefficient Standard t-Statistic p-Value on Gender Error 20.0029 0.0034 20.88 0.38

A similar study was conducted for the WNBA,83 which has existed since 1997. From 1997 to 2015, there have been 56 coaches who coached at least one entire season. Of these, 29 have been men and 27 women. Thus, 48% of coaches in the WNBA are women. Given the slight preference for men in the WNBA, and the overwhelming preference for men in the NBA, we again wonder if male coaches have a larger impact on individual player performance.

81This is measured following the work of Berri and Krautmann (2013). 82The model, based on the work of Darvin et al. (2016), employed conference fixed effects. It also only considered players who played at least 200 minutes in consecutive seasons. Overall, 3,615 observations were employed. 83Darvin et al. (2016).

09_BER_46412_ch08_259-290_PP3.indd 280 8/31/17 1:04 PM Women and Sports 281

To answer this question, the factors listed in Table 8.9 were utilized in a model designed to see if the gender of a coach impacts the productivity of individual players. 84

Table 8.9 Evaluating Gender and Coaching in the Women’s National Basketball Association Dependent Variable PROD40: A player’s overall productivity per 40 minutes played84 Independent Variables Gender: dummy variable for gender of coach (1 = female, 0 = male) Age: age of the player (squared term is included because this is not expected to be a linear relationship) GP: games played last two seasons (to control for injury) TMPROD: productivity of teammates NewTeam: dummy variable for moving to a new team NewCoach: dummy variable for moving to a new coach DBIG: dummy variable equal to 1 if the player is a power forward or center DGUARD: dummy variable equal to 1 if the player is a point guard or shooting guard

To estimate the model, data for each WNBA player were collected for 19 seasons, beginning with 1997 and ending in 2015.85 Once again, as Table 8.10 details, the story seems to be the same as we saw in the study of women’s college basketball. The gender of a coach in the WNBA is not related to player performance.86

Table 8.10 Link Between Gender of Coach and Individual Player Performance in the Women’s National Basketball Association86 Coefficient on Gender Standard Error t-Statistic p-Value 20.0031 0.0043 20.71 0.48

84This is measured following the work of Berri (2008) and Berri and Schmidt (2010). 85Data for each player were obtained through online databases such as archived team rosters, archived player profiles, and http://www.basketball-reference.com/wnba/. The model also included player fixed effects to control for characteristics such as the quality of the player. 86The model, based on the work of Darvin et al. (2016), employed player fixed effects. It also only considered players who played at least 200 minutes in consecutive seasons. Overall, 1,620 observations were employed.

09_BER_46412_ch08_259-290_PP3.indd 281 8/31/17 1:04 PM 282 Sports Economics

So what have we learned? The market appears to suggest that being a man is helpful. And all these results make it very clear that having a male coach does not lead to higher player productivity. We would prefer to see a similar scenario for the NBA. But in 2015‒16, only two women, Becky Hammon and Nancy Lieberman, worked as coaches in that league. And both are only assistant coaches. Once again, there has never been a woman hired as a head coach in a major North American professional sport.87 Women are clearly underrepresented in every sector of the economy. But women do work as leaders of Fortune 500 companies, university presidents, gov- ernors of states, members of the U.S. Congress, and heads of nations. If women can lead in all these other settings, why are sports teams not hiring them to lead?88 One explanation sometimes offered is that women do not respond well to pressure. A paper by Cohen-Zada et al. (2016) cast doubt on such an explanation.89 These authors looked at data from the four Grand Slam tennis tournaments of 2010. With a data set consisting of 4,127 tennis games for women and 4,153 games for men, they found that men consistently perform worse as competitive pressure increases.90 In sum, men are more likely to choke. This same result was observed even when the authors considered a variety of different empirical approaches (i.e., the evidence appears to be robust).91

87Nancy Leiberman did coach an NBA Development League team from 2009 to 2011. And Becky Hammon coached the San Antonio Spurs summer league team in the summer of 2015 and 2016. Leiberman’s team made the playoffs in 2011, while Hammon’s team won the Las Vegas summer league title in 2015 (http://www.newsday.com/sports/columnists/barbara-barker/ mike-francesa-s-rant-about-women-coaches-reflects-stone-age-thinking-1.13209552). 88If you talk to women who are sports fans, they often cite a common misperception. Men frequently question whether or not women truly understand sports—an observation shared by Molly Cosby (http://www.theladiesleague.org/single-post/2016/1/25/Should-Women-Go- Through-Stricter-Security-Screenings-to-Enter-Sporting-Events). Such an explanation both (1) greatly exaggerates how complicated sports might be to understand and (2) insults women. If you are a male who resorts to it . . . you really need to stop! 89Danny Cohen-Zada, Alex Krumer, Mosi Rosenboim, and Offer Moshe Shapir, “Choking Under Pressure and Gender,” working paper, October 2016, https://www.researchgate.net/ publication/308901292_Choking_Under_Pressure_and_Gender. 90The authors of this study note that the tennis matches examined were not mixed. They there- fore suggest it is possible their results do not extend to competitive circumstances, where both men and women are participating at the same time. 91The authors note there may be a biological reason for this outcome. There is a literature that suggests cortisol levels increase more rapidly for men as pressures mount. Cortisol has been found to harm a mind’s critical abilities.

09_BER_46412_ch08_259-290_PP3.indd 282 8/31/17 1:04 PM Women and Sports 283

The study of tennis suggests that when it comes to choosing leaders, the issue isn’t how men and women respond to competition. A more likely explanation is that men are favored because women face discrimination. And there is a cost to such discrimination. As we will observe in Chapter 11, although most coaches do not appear to be capable of altering player performance, some have proved able to do this. As we noted in our discussion of discrimination, a team is less likely to find the best talent if it does not consider all qualified candidates.

8.5 To Understand Gender and Sports, You Need to Look Beyond Markets

Economists tend to focus on market forces as the explanation of the outcomes observed in the economy. And sometimes that focus is correct. But our discussion of women and sports has revealed we frequently need to look beyond the market to explain what we are observing. Nearly 100 years ago, women’s soccer was a popular sport in England, attracting thousands of fans. Then the Fédération Internationale de Football Association (FIFA), an organization led by men, shut it down.92 Decades later, it was believed women simply didn’t like sports. And then Title IX was passed in the United States, forcing schools to offer sports programs to women. Suddenly, we learned women did indeed enjoy playing and watching sports. These stories highlight how non-market forces can impact outcomes. We encountered a similar story when we turned our attention to the compensation of athletes and assignment of leadership positions. Again and again, we have to do more than say, “This is simply a reflection of supply and demand.”

92Today, women do participate in FIFA. But a recent vote illustrates that men still control the organization. As Julie Foudy of ESPN.com reported, the Asian Football Confederation held a vote in May of 2017 to fill a position for a female member of the FIFA Council. Four candidates ran for the position. The person most widely considered the most qualified was Moya Dodd, a woman and former player for the Australian National Team. The woman selected was Mahfuza Akhter Kiron. 27 of the 44 voters (all male) selected a woman who later revealed she did not know who won the last Women’s World Cup in soccer. If one does not know the very basic facts about the sport one is leading, one is not likely to be able to do much to reform the sport. But this is who an all-male group decided was the best woman to help lead FIFA. As Foudy argued, such a vote doesn’t exactly represent the best approach to equality in soccer. http://www.espn.com/espnw/voices/article/19364609/ dear-fifa-do-better-support-equal-representation

09_BER_46412_ch08_259-290_PP3.indd 283 8/31/17 1:04 PM 284 Sports Economics

Problems

Study Questions 1. What is Title IX, and when was it enacted?? 2. How has the number of girls/women playing high school/college sports changed from before the passage of Title IX to today? What role have so-called market and non-market forces played in the participation rates? 3. How popular was women’s soccer in England prior to 1921? How and why did this change in 1921? 4. How is FIP ERA calculated? Why do we believe this is a better measure of a pitcher’s performance? 5. How does demand for the WNBA today compare to demand for the NBA today? How does demand for the WNBA today compare to what we saw in the NBA after 20 years? 6. According to Cooky, Messner, and Musto (2015), how does coverage of women’s sports compare to that of men’s sports? How has this changed over time? 7. How was the gender wage gap in professional tennis overcome? 8. How does the percentage of revenue paid to WNBA players compare to what we see for the NBA? What explains the difference in the two leagues? 9. In general, what percentage of leadership positions goes to men in the economy? 10. What role does the gender of a coach play in outcomes for college softball, women’s college basketball, and the WNBA? Does the evidence support the idea that men are better leaders? 11. Who is better in the “clutch,” men or women? Reference the study of Grand Slam tennis in answering this question. 12. We noted in the text that attendance is relatively low early in a league’s time span. Design a model to predict attendance in a league. List and explain all independent variables that you believe might explain attendance. 13. Demand for college women’s sports and international competition involving women’s teams seems much higher than what we for professional women’s sports leagues. What might explain this difference?

Appendix 8A Econometric Models for Women’s Sports

The chapter reviewed several empirical models designed to study gender and sports. The first was our model to estimate FIP ERA for pitchers in the NPF. Table 8A.1 presents the results of that model’s estimation.

09_BER_46412_ch08_259-290_PP3.indd 284 8/31/17 1:04 PM Women and Sports 285

Table 8A.1 Dependent Variable as a Pitcher’s ERA (minimum 50 innings pitched): 2004–1693 Variable Coefficient t-Statistic p-Value Strikeouts per 7 innings pitched 20.123 26.30 0.000 Walks per 7 innings pitched 0.195 4.89 0.000 Hit-by-pitch per 7 innings pitched 0.478 3.82 0.000 Home runs per 7 innings pitched 1.623 14.60 0.000 Hits per ball in play (per 7 innings pitched) 12.729 12.13 0.000 Constant 21.916 25.62 0.000 Observations 191 R-squared 0.87

The second model came from the work of von Allmen (2013). This study estimated the following model: 93

RPI 5 a0 1 a1 3 ShareExp 1 a2 3 WS10 1 a3 3 PAC-10 1 a4 3 GENDER (8A.1) where RPI 5 rank of team according to the rating percentage index (an index that considers both winning percentage and strength of schedule) ShareExp 5 ratio of expenditures of each program to the expenses of all programs in the sample WS10 5 number of World Series appearances over the 10-year period, 1996‒2005 PAC-10 5 dummy variable for teams in the PAC-10 (the dominant conference in softball) GENDER 5 du mmy variable for gender of coach (1 5 female, 0 5 male) The estimation of this model is reported in Table 8A.2. As one can see, the model indicates that success in softball is linked to spending and past success. Such a result is consistent with what we will note in the next chapter. Success in college sports is heavily linked to past success, since it appears a school’s past success in a sport influences the choices of today’s recruits. The third and fourth model both looked at how the gender of a coach impacts the performance of an individual player. We first examined this issue with respect to women’s college basketball. The specific model94 is detailed in equation 8A.2:

PROD40 5 b0 1 b1 3 DGENDER 1 b2 3 lagPROD40 1 b3 3 CLASS 1 b4 3 GM2 1 b5 3 TOTEXPPER 1 et (8A.2)

93Data for this estimation derive from http://www.profastpitch.com/home/. 94The model detailed in equation 8A.2 is derived from Darvin et al. (2016).

09_BER_46412_ch08_259-290_PP3.indd 285 8/31/17 1:04 PM 286 Sports Economics

Table 8A.2 Explaining Success in College Softball95 Variable Coefficient Standard Error t-Statistic p-Value Constant 92.01 7.91 11.63 0.000 Gender 4.14 5.17 0.80 0.425 ShareExp 23,933.81 650.44 26.05 0.000 WS10 23.11 1.58 21.97 0.052 PAC-10 29.56 10.78 20.89 0.378 Adjusted R-Squared 0.41

where PROD40 5 player productivity, per 40 minutes95 DGENDER 5 dummy variable for gender of coach (1 = woman, 0 = man) lagPROD40 5 PROD40, last season CLASS 5 equal to 1 for freshmen, 2 for sophomore, etc. GM2 5 games played last two seasons (control for injury) TOTEXPPER 5 total expenditure per player for team96

The model was estimated with conference fixed effects.97 The results are reported in Table 8A.3.98

Table 8A.3 Explaining Individual Performance in College Women’s Basketball: 2012–13 to 2014–1598 Independent Standard Coefficient t-Statistic p-Value Variables Error Gender 20.0029 0.0034 20.88 0.38 Class 0.0041 0.0019 2.13 0.03 GP 0.0023 0.0003 7.57 0.00 TMPROD 20.0166 0.0464 20.36 0.72 DBig 0.0188 0.0043 4.40 0.00 DGuard 20.0030 0.0045 20.66 0.51

(continued)

95von Allmen (2013, p. 281). 96These data come from the Department of Education (http://ope.ed.gov/athletics/Index.aspx). 97Fixed effects are dummy variables employed in panel data. As Peter Kennedy noted: “The dummy variable coefficient reflect ignorance—they are inserted merely for the purpose of measuring shifts in the regression line arising from unknown variables” (p. 222). A Guide to Econo- metrics, 3rd ed. (Cambridge, MA: MIT Press, 1996). 98This model is derived from the study reported in Darvin et al. (2016).

09_BER_46412_ch08_259-290_PP3.indd 286 8/31/17 1:04 PM Women and Sports 287

Table 8A.3 ( continued) Independent Standard Coefficient t-Statistic p-Value Variables Error TotExp 2.95E-08 3.72E-08 0.79 0.43 LagProd 0.5042 0.0191 26.34 0.00 Constant 20.0816 0.0175 24.67 0.00 R-squared 0.25 Adj. R-squared 0.24 Observations 3,615 Minimum minutes 200

The model was estimated with data from the 2012‒13, 2013‒14, and 2014‒15 seasons. Although men’s college basketball data exist back to 2002‒03, women’s data are less plentiful. ESPN.com does not even report data from the current season. The NCAA reports women’s data back to 2009‒10, but the minutes data are only correct from 2012‒13 to the present.99 This thus limits the study to some extent. Despite this issue, the model — as detailed in the chapter — did suggest that the gender of a coach is not related to player performance. It also tells a few additional stories:

•• Players do improve throughout their time in school, and the more games a player participates in, the better she performs. Obviously, this latter result may also indicate that better players get more games. •• It was hypothesized that player performance could be related to spending on players. The model indicates that such is not the case. This suggests team spending does not typically alter on-court productivity. One should note that conference fixed effects are included and these are related to team spending. So that might help explain the result observed with respect to spending.

The third model detailed in the chapter looked at coaching in the WNBA. The specific model100 designed to explain productivity per 40 minutes in the WNBA101 is noted in equation 8A.3:

99http://stats.ncaa.org/team/inst_team_list?sport_code=WBB&division=1. 100This model is based on the Darvin et al. (2016). 101This is measured following the methodology detailed in Berri and Krautmann (2013) without the adjustment made for position played. In other words, this is not Wins Produced (which includes an adjustment for position played). The position adjustment is necessary if you are

09_BER_46412_ch08_259-290_PP3.indd 287 8/31/17 1:04 PM 288 Sports Economics

PROD40 5 b0 1 b1 3 DGENDER 1 b2 3 AGE 1 b3 3 SQAGE 1 b4 3 GM2 1 b5 3 TMPROD40 1 b6 3 DNEWTEAM 1 b6 2 DNEWCOACH 1 b8 3 DBIG 1 b9 3 DGUARD (8A.3)

where PROD40 5 player productivity, per 40 minutes DGENDER 5 dummy variable for gender of coach (1 = woman, 0 = man) AGE 5 age of player GM2 5 games played last two seasons (control for injury) TMPROD40 5 teammate productivity per 40 minutes DNEWTEAM 5 dummy variable for new team DNEWCOACH 5 dummy variable for new coach DBIG 5 dummy variable for center or power forward DGUARD 5 dummy variable for point guard and shooting guard

The model was estimated with player-specific fixed effects. The results, which are for players with at least 200 minutes in the current and past season, are reported in Table 8A.4.102

Table 8A.4 Explaining Individual Performance in the Women’s National Basketball Association: 1998–2015102 Independent Standard Coefficient t-Statistic p-Value Variables Error Gender 20.0031 0.0043 20.71 0.48 Age 0.0129 0.0071 1.82 0.07 Age, Sq. 20.0003 0.0001 22.12 0.03 GM2 0.0011 0.0004 2.97 0.00 TMPROD 20.0941 0.0682 21.38 0.17 DNewTeam 0.0033 0.0053 0.62 0.53 DNewCoach 20.0141 0.0045 23.11 0.00 DBig 0.0156 0.0102 1.52 0.13 DGuard 20.0178 0.0095 21.86 0.06 Constant 20.0466 0.1077 20.43 0.67 (continued)

comparing players at different positions. But in a study like this, you are only looking at how the performance of a specific player changes. In addition, the position adjustment is only an approximation. Clearly, some players are centers and some are guards. But where a specific player ends up on that spectrum is somewhat arbitrary and can change from year to year. Hence, as was shown in Berri et al. (2009), the position adjustment is not included. 102This study is based on the work reported in Darvin et al. (2016).

09_BER_46412_ch08_259-290_PP3.indd 288 8/31/17 1:04 PM Women and Sports 289

Table 8A.4 ( continued) Independent Standard Coefficient t-Statistic p-Value Variables Error R-squared 0.74 Adj. R-squared 0.65 Observations 1,620 Minimum minutes 200

In Chapter 8, we discussed the results with respect to gender. Here are some other stories told by this model:

•• Both age and age squared are statistically significant and of the correct signs. The results indicate that a WNBA player tends to peak at 25 years of age. In addition, Table 8A.5 reveals how age generally impacts performance. If a player is average at her peak (at 25 years of age), then she will likely not dip below the 0.090 mark until 32 years of age. Such a result reveals that WNBA players appear to be less impacted by age than NBA players.

Table 8A.5 Impact of Age on Women’s National Basketball Association Players103 Age WP48 Age WP48 19 0.091 27 0.099 20 0.094 28 0.098 21 0.096 29 0.096 22 0.098 30 0.094 23 0.099 31 0.091 24 0.0997 32 0.087 25 0.1000 33 0.083 26 0.0997 34 0.079

•• The results suggest that playing more games leads to higher productivity. Obviously, this can reflect the choices of coaches. But it also likely reflects the impact of injury. Ideally, we would include actual data on injuries.103 •• Although the gender of a coach did not matter, moving to a new coach generally saw a dip in productivity. This may indicate that it takes time for players to adjust to a new system.

103This table reports expected performance for a player who is average at the age of 25, or the estimated peak age for WNBA players.

09_BER_46412_ch08_259-290_PP3.indd 290 8/31/17 1:04 PM